Peer-graded Assignment: Programming task
Here is your final project!
As we discussed in the video, we propose you to build a model of House Pricing dataset with the mean squared error as a loss function. In your task you should manage to calculate loss’s gradient and write a simple gradient descent. As a creative part of the task, experiment with the step length and initial weights.
All instructions are provided in Jupiter Notebook attached. Please, also download the rest of the files xtrain.npy, ytrain.npy and checker; those are datasets and automatical checker.
Do not alter auxiliary files! Place them in the same directory as ipynb. You need to work only with the Notebook file.
In case you don’t have a local Python Installation, you can use Google Collaboratory colab.research.google.com
Review criteria
| Task | Criterion | Points |
|---|---|---|
| Task 1. Reading and Preparing. | Review cell with checker for this task. | 0,5 |
| Task 2. Compute analytically the function. | Review cell with checker for this task. | 1 |
| Task 3. Compute analytically the gradient of the . | Check correctness of the computed gradient. | 1 |
| Task 4. Write a function to compute the gradient of the Loss function in the given point. | Review cell with checker for this task. | 1 |
| Task 5. Write gradient descent. | Check for at least 2 experiments. Check if at least one graph converges. | 1 |
| Task 6. Discussion. | Check answers. | 1,5 |